Cognitive and gait deficits are common symptoms in Parkinson’s disease (PD) that worsen with disease progression (1). Motor-cognitive dual tasks (DTs) (e.g., walking while talking) as a method to investigate the interplay between gait and cognition require both motor control and cognition to perform the DT (2, 3). It is well known that DTs affect gait even in healthy individuals by reducing gait velocity (4, 5). The interference between motor control and cognitive demand is important for motor-impaired patients in everyday life activities. In mild to moderately affected PD patients, cognitive DTs during walking (subtracting numbers, digit span DT) reduced gait performance and increased gait variability (4, 6–8). This indicates that dual tasking may be responsible for trip and fall risk in patients with motor impairments due to additionally challenges on motor control (9–11).

Walking is a complex task in which cognitive resources continuously monitor bilateral coordination and dynamic postural control, both necessary for the walking process including cognitive-motor control (3). In this context, a systematic review revealed that there is a direct relationship between cognitive deficit severity and gait abnormalities in patients with dementia (12). Cognitive impairment in PD typically consists of deficits in attention, executive and visuospatial functions, as well as memory resources (13). Even in early PD, attention and executive function deficits are features of basal ganglia pathology (14) and necessary to appropriately allocate cognitive resources for the optimal performance of simultaneous tasks (15). Therefore, it is not surprising that attention and executive function are associated with DT performance (4, 16, 17).

In clinical routine the Montreal Cognitive Assessment (MoCA) is a widely used and international accepted scale to evaluate the cognitive status in PD (18–22). It is a 30-point scale that evaluates visuospatial function, language, digit span, executive functions, attention, and memory (23). It has been previously shown that executive functions, attention, visuospatial abilities, and memory are associated with gait impairment in PD (3, 5, 15, 24, 25). In particular, the association between executive functions and DT gait performance has been described (4, 6, 15). Evaluating DT situations in PD is useful since cognitive functions and motor control—both often impaired in PD—are examined at the same time. DT costs are a common measure to evaluate the effect of the secondary (cognitive) task on the primary task (walking) and have been demonstrated to reflect impaired DT gait performance in PD (5).

With regard to the interplay between gait, cognition, and dual tasking, the majority of studies assessed the effect of DT on gait or the effect of cognitive deficits on gait (5, 6, 16, 25). However, it is not precisely characterized whether cognitive performance can be predicted by gait performance under DT condition. One study observed that DT affects swing time, step length variability, and single/double support time ratio more pronounced in PD patients with mild cognitive impairment compared with controls (24). This indicates that gait might serve as an objective clinical biomarker for cognitive gait control which could support fall risk detection and fall prevention in PD. Since cognitive deficits are correlated with a larger fall risk (26) and associated with a shorter survival time (27, 28) it is worth studying if cognitive deficits may be mirrored by gait performance to support the clinical routine with quantitative data and early detect risk factors for falls.

The aim of this study was to investigate if cognitive performance evaluated with a clinical cognitive assessment (MoCA) may be reflected by DT costs of spatiotemporal gait parameters. In this context it is interesting to detect which DT gait parameters are sensitive to cognitive impairment.

We hypothesize that cognitive deficits correlate with DT gait performance in PD patients and that cognitive deficits can be predicted by DT gait parameters. To test this hypothesis, three analyses were performed: (1) correlation analysis between MoCA score and DT costs for each gait parameter separately [DT costs are defined as the percentage change between single task (ST) and DT gait parameters], (2) linear regression analysis that includes most relevant gait parameters (based on the correlations) and confounders of DT gait performance (e.g., age and disease duration), and (3) a comparison of cognitively impaired and unimpaired PD patients in two groups that are matched by clinical confounders. In summary, we observed subtle correlations between cognitive performance and DT gait costs. Cognitively impaired and unimpaired PD patients were equally challenged by the DT gait test.

This study investigated the interplay of cognitive deficits in PD and gait performance during a widely used DT paradigm. We hypothesized that PD patients with cognitive deficits show larger DT costs in gait parameters. However, our results did not reveal correlations between DT costs of gait parameters and cognitive performance assessed with MoCA. DT gait costs combined with clinical confounders explained about 30% of the MoCA score variance. Comparing matched groups of cognitively impaired and unimpaired PD patients, we did not find significant differences in DT gait costs. Confounder effects on DT walking as previously described (17, 24) were excluded since we matched groups by clinical confounders such as age, gender, disease duration, H&Y, UPDRS-III, medication and depressive symptoms. These findings suggest that DT gait performance is not indicative for cognitive impairment in PD assessed with MoCA. There are several considerations to be drawn based on the present finding.

Cognitive performance of PD patients was particularly affected in the MoCA subdomains visuospatial/executive function, attention, language, abstraction, and memory. The performance in these subdomains is consistent with a larger PD cohort (n = 486) (44). Interestingly, in our study cognitive impairment is inadequately reflected by DT gait costs. The cognitively impaired group reached a significant lower score in the MoCA task “serial 7 subtraction” compared with the unimpaired group. It has been shown that serial subtraction in steps of 3 and steps of 7 while walking has a substantial effect on gait in PD (6). Since serial 7 subtraction is more challenging than the DT “serial 3 subtraction,” the comparability of the cognitive performance between MoCA task and the cognitive task of the DT is limited. However, even the less challenging serial 3 subtraction substantially distracted the patients and aggravated the gait impairment (large effect sizes for DT on gait parameters). Intriguingly, this effect was present in both groups, e.g., independently from the cognitive performance assessed by MoCA. This suggests that motor-cognitive DT during walking may not be sensitive enough to detect mild cognitive impairment in PD patients. Additional investigations are needed to explore the role of different challenging motor-cognitive DTs to address their sensitivity to detect mild cognitive impairment in PD.

A direct relationship between cognitive deficits and gait impairment has been described in patients with dementia (12). Attention and executive function deficits are associated with basal ganglia pathology (14). Both cognitive domains are impaired in PD (45) and necessary to appropriately allocate cognitive resources for gait control in single and DTs (15, 46). We used the MoCA score in this study since this test adequately detects mild cognitive impairment in PD (35). Significant differences in DT gait costs could not be detected although matched groups significantly differed in the MoCA subsections “executive function” and “attention.” However, limits of the MoCA in detecting executive function and attention that are relevant for motor control should be taken into account. Impaired executive functions assessed by a trail making test have been shown to negatively influence DT gait costs in PD (6). In the MoCA, a short version of trail making is also included but maybe executive functions necessary for motor control during DT may not be sufficiently detected with this version. The MoCA as global cognitive assessment used in the clinical routine may not be as precise as extensive neuropsychological test batteries (47, 48). In a recent study using a neuropsychological test battery including trail making for executive function, verbal episodic memory, naming, attention, and working memory tests, DT gait has been shown to correlate with progression in dementia in older adults (n = 112) with mild cognitive impairment (49). This indicates on the one hand that more precise assessments might be necessary to detect cognitive deficits specific for DT gait. On the other hand, patients with progressed dementia were particularly impaired in naming and attention but not in executive functions indicating that other parts of cognitive-motor control are affected compared with PD patients. In general, it should be taken into account that the MoCA may be a global cognitive assessment tool but not sensitive enough to detect executive functions responsible for DT gait performance in PD.

PD patients were not instructed to prioritize walking or counting during the DT. In this study, DT costs for gait parameters were recorded, but DT costs for the cognitive task (incorrect answers in counting backwards) were not covered. Therefore, it may be possible that PD patients differently prioritized cognitive and motor task. Bloem et al. (50, 51) reported that PD patients may use the so called “posture second strategy,” where they focus less on maintaining a safe gait compared with healthy controls. We observed similar gait performance in both groups during DT walking and found descriptively larger DT costs of gait velocity in the cognitively unimpaired group which may indicate that this group focused more on the cognitive task. In this context, it has also been described that motor-cognitive DTs show twofold effects on gait performance. They reflect a PD-independent, age-related reduction in gait parameters, and a PD-specific coordination deficit during DT that affects postural control (52). Smulders et al. (53) observed that DT performance in the gait or cognitive task does not correlate with prospective falls in a large PD cohort (n = 263). This contradicts previous studies which observed associations between fall risk and DT gait (9–11). Taken together, in the light of these studies our results suggest that DT gait performance should be carefully interpreted in the clinical context due to variable individual responses under this challenge. DTs have the potential to reveal motor-cognitive deficits that are linked to basal ganglia pathology; however, appropriate assessments of gait and cognitive performance are necessary, and influencing factors should be controlled. Future studies should strictly control task prioritization by recording DT costs for gait and cognition.

Correlation analysis did not reveal significant correlations between MoCA score and DT gait costs. However, DT costs of stride length, gait velocity, swing time CV, and max TC showed at least weak correlations to cognitive impairment in this cohort and have previously been described as parameters that are affected by DT walking in PD (6–8). Future studies should focus on DT costs of those gait parameters and evaluate the predictive value in terms of cognitive impairment. Max TC may be an interesting parameter with regard to fall risk in PD during DT. Trip risk and falls in cognitively impaired PD patients are not well understood due to the fact that PD patients with cognitive impairment are often excluded in fall-related studies in PD (54). Further research should focus on the relevance of foot clearance to detect trip risk during DT walking in cognitively impaired PD patients.

The results of this study indicate that DT gait performance did not reflect cognitive impairment in PD assessed by MoCA. Instead, cognitively impaired and unimpaired PD patients were equally challenged by the DT. Differences in task prioritization and less sensitive cognitive assessment using MoCA to evaluate executive functions specific for DT gait control may have influenced the results. Furthermore, DT gait parameters should be carefully interpreted in the clinical context due to variable individual responses to DTs.

The study was approved by the local ethics committee (IRB-approval-Re. No. 4208, IRB, Medical Faculty, Friedrich-Alexander University Erlangen-Nürnberg, Germany), and all participants gave written informed consent according to the Declaration of Helsinki.

HG was responsible for conception, design, and organization of the study. He performed data acquisition, statistical analysis, and interpretation, and wrote the manuscript. FM, JS, and ZK supported data acquisition and reviewed the manuscript. BE, WA, KP, and SS collaborated in this project, critically discussed findings, and reviewed the manuscript from a technical, statistical, and movement science perspective. JW and JK supported in designing the study, critical revision, and preparing the manuscript.

HG, KP, JS, ZK, WA, and SS declare no conflict of interest. FM is supported by the Interdisciplinary Center For Clinical Research (IZKF) of the University Hospital Erlangen. BE holds ownerships of Portabiles HealthCare Technologies GmbH and Portabiles GmbH, received compensation and honoraria from serving on scientific advisory boards for Abbvie GmbH, adidas GmbH, Bosch Sensortec GmbH, and ST Sportservice GmbH. JW reports personal fees outside of the submitted work from Teva GmbH, Ever Pharma GmbH, Desitin Arzneimittel GmbH, Abbvie GmbH & Co. KG, Biogen GmbH, and GlaxoSmithKline GmbH & Co. KG. JK holds ownerships of Portabiles HealthCare Technologies GmbH and Portabiles GmbH, received compensation and honoraria from serving on scientific advisory boards for LicherMT GmbH, Abbvie GmbH, UCB Pharma GmbH, GlaxoSmithKline GmbH & Co. KG, Athenion GmbH, and Thomashilfen GmbH; as well as lecturing from UCB Pharma GmbH, TEVA Pharma GmbH, Licher MT GmbH, Desitin GmbH, Abbvie GmbH, Solvay Pharmaceuticals, and Ever Neuro Pharma GmbH.